Re: [AMBER] GAMD/TI issue in Ambaer20/22

From: mish via AMBER <amber.ambermd.org>
Date: Wed, 24 May 2023 17:36:08 -0500

...and with Amber 20 (as I heard that there are some issues with gamd in
amber22), I can see that a well-equilibrated system breaks apart within
2 steps of gamd:
https://www.dropbox.com/s/4animxsln3fu0uq/gamd_amber20_err.jpeg?dl=0

 Whatever ligand I am giving in timask1 and scmask1 starts distorting
first, and then slowly, the whole system becomes unstable. At this point, I
feel something is wrong with the game or TI code.

Best,
Mish


On Wed, May 24, 2023 at 4:54 PM mish <smncbr.gmail.com> wrote:

> Many thanks. I have repeated simulations with iwrap=0 in the past and just
> now, but the issue remains the same:
> https://www.dropbox.com/s/13cgyo3legiimlc/Gamd_err1.png?dl=0
>
> Suddenly, two ligands start distorting within a few hundred steps, and I
> still do not understand what's going wrong. cMD works fine, and I am
> starting these game simulations using the restart file from 1-microsecond
> cMD in pmemd.cuda. I have also tried up to 100ns cMD using CPU code and
> then restarting gamd but the issue remains the same.
>
> Do you think there is any mistake I am making with the MD input setting?
>
> &cntrl
> imin=0, irest=1, ntx=5,
> ntt=3, gamma_ln=1.0, temp0=300,
> ntpr=50, ntwx=50, ntwr=50000,
> cut=9.0, nstlim=500000, dt=0.001,
> ntc=1, ntf=1,
> iwrap=0, barostat=2,
> ntp=1, pres0=1.0,
> ntxo=2, ioutfm=1,
> igamd = 11, irest_gamd = 0,
> ntcmd = 700000, nteb = 27300000, ntave = 140000,
> ntcmdprep = 280000, ntebprep = 280000,
> sigma0P = 4.0, sigma0D = 6.0, iEP = 2, iED=1,
> icfe = 1, ifsc = 1, gti_cpu_output = 0, gti_add_sc = 1,
> timask1 = ':1-2', scmask1 = ':1-2',
> timask2 = '', scmask2 = '',
> ibblig = 1, nlig = 10, atom_p = 667, atom_l = 19, dblig = 3.7,
> &end
>
>
> Mish
>
> On Wed, May 24, 2023 at 1:00 PM David A Case <david.case.rutgers.edu>
> wrote:
>
>> On Wed, May 24, 2023, mish via AMBER wrote:
>>
>> >The problem seems to occur when one of the free ligands is close to the
>> >boundary, and wrapping it back in the box completely messes up the
>> system.
>> >Any suggestions on how to prevent it?
>>
>> It's certainly worth setting iwrap to the default value of zero. If you
>> look at the description of this variable (section 21.6.3 in the Reference
>> Manual), you'll see that there are very few (modern) circumstances in
>> which
>> a non-zero value is useful. (If you *do* need to wrap the simulation
>> into a
>> periodic box, say for visualization, it is better to use cpptraj for that
>> purpose, since cpptraj is much more flexible.)
>>
>> It is possible (but outside my bailiwick) that ligand GaMD does is not
>> fully
>> periodic. But take care, especially if you have some reason for setting
>> iwrap=1 in the first place.
>>
>> ....dac
>>
>>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Wed May 24 2023 - 16:00:02 PDT
Custom Search